Global Hypothesis Verification on UWA

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

Global Hypothesis Verification on UWA

zoltan.rozsa
Hello,

I tried to test the Global Hypothesis Verification sample on UWA database (converting the ply files to pcd fromat with CloudCompare program), but it did not find any good result. If I know well it was tested on this database with 100 % RR. What options should I set to get correct results (with the provided milk.pcd and milk_cartoon_all_small_clorox.pcd files it works fine)?

Thank you

_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Global Hypothesis Verification on UWA

andersgb1
The original publication uses multiple rendered views of the models for matching - it also lists all the parameter settings. So I think you need to do all of this to make it work that well.

On 04/14/2015 04:39 PM, [hidden email] wrote:
Hello,

I tried to test the Global Hypothesis Verification sample on UWA database (converting the ply files to pcd fromat with CloudCompare program), but it did not find any good result. If I know well it was tested on this database with 100 % RR. What options should I set to get correct results (with the provided milk.pcd and milk_cartoon_all_small_clorox.pcd files it works fine)?

Thank you


_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users


_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Global Hypothesis Verification on UWA

Ber461
Strangely, when I run the (unchanged) tutorial code using the provided files (milk*.pcd), I get no results. This is very strange, as the default parameters were fine tuned by the authors for those same files. Anyone else has this problem?
Reply | Threaded
Open this post in threaded view
|

Re: Global Hypothesis Verification on UWA

lilouch
Yeah same here. Did you able to make it work ?
Reply | Threaded
Open this post in threaded view
|

Re: Global Hypothesis Verification on UWA

mczhouwei
This post has NOT been accepted by the mailing list yet.
In reply to this post by andersgb1
I also meet this problem, firstly I try SHOT descriptor with Global HV method, and use the parameter mentioned in the eccv 2012 paper and also tried the parameter mentioned in PCL tutorial, totally can only find bad instance, no any good instance.
Even I try with other dataset with other descriptor, in my opinion I think I get a very precise instance, but even this it also say it's a bad instance.
And I read this paper several times, and view some little about the sorce code with Gloabl HV, I guess the problem maybe is in Simulated Annealing parameter setting, the source problem is the threthold set for cost function, also can say how to distinguish an instance is good or bad?